TB National Survey in Indonesia

TB National Survey in Indonesia

Muhammad N. Farid Workshop on repeat prevalence surveys in Asia: design and analysis 8 - 11 February 2012 - Phnom Penh, Cambodia

TB Related National Survey

TB Prevalence Survey 2004

Basic Health Research 2007

Basic Health Research 2010

TB Prevalence Survey 2012 (planning)

TB Prevalence Survey 2004

Survey Design Piggy backed to National Socio-Economic Survey

(250,000 HHs)

Two-stage cluster stratified sampling design (stratified by urban/rural area)

Census block as PSU (select by PPS) and Household as SSU (16 HHs per CB)

Sample size of 58,000 adults (20,000 HHs) covered 30 provinces

National and regional level of estimation

Sampling Design

Population Sampling frame of census

block

NSES-Core Samples (15,625 CBs or 250,000 HHs)

NSES-Module Samples (4,200 CBs or 67,000 HHs)

TB Prevalence Survey (1,250 CBs or 20,000 HHs)

Select 16 HHs per CB systematically

Suspect Definition

Person who had a productive cough, yielding sputum and/or blood, of any duration at any time during the preceding month

Case definition of smear-positive pulmonary TB case

At least 2 (out of 3, i.e., spot, morning, spot when collect morning sputum) sputum specimens smear-positive for acid-fast bacilli by microscopy

Finding suspects and sputum collection

Start Visit 1250 CBs and

sampled 20,000 HHs

Screening interview to find TB suspects for all HH member

aged 15yr or older

Find suspects

?

Collect sputum (3 times)

Next HH/CB?

Stop

Culture (45% of samples) and microscopic examination

Yes

Yes

No No

Sample size

Type Plan Visited/inter

viewed Response

rate

Household 20,000 HHs 18,387 HHs 92.0%

KAP respondent 20,000 HHs 17,887 HHs 89.4%

HH members <15yrs 26,335 22,551 85.6%

HH members 15yrs+ 57,818 50,154 86.7%

TB Prevalence Survey 2012

Survey Design Two-stage cluster stratified sampling design

(stratified by region, i.e., Sumatera, Java-Bali, and Others)

Village as PSU (select by PPS) and group of CB as SSU (1 group of CB per village and 500 adults per group of CB)

Sample size of 78,000 adults (156 clusters) covered 33 provinces

National and regional level of estimation

Sample Size

Estimate of TB prevalence per 100,000 (p) based on WHO estimates: 156

Deff = 1.5

Cluster size (M) = 500

Precision (d) = 20%

Response rate (r) = 85%

Calculated sample size:

Adults = 78,000

Clusters = 156

Sampling Design

Population Sampling frame of village

(>70,000)

Select 156 villages by PPS sampling, stratified by region

Develop sampling frame of cluster (group of CBs)

Select 1 cluster per selected village by PPS sampling

1 cluster consists about 500 adults

Suspect Definition

Person who had a productive cough for at least two weeks and/or an abnormal chest X-ray

Case definition of smear-positive pulmonary TB case 2 sputum specimens smear-positive for acid-fast

bacilli by microscopy, or

At least 1 sputum specimen smear-positive for acid-fast bacilli by microscope, radiographic abnormalities consistent with pulmonary TB, and a decision of a physician to treat with the full course of chemotherapy, or

At least 1 sputum specimen smear-positive for acid-fast bacilli by microscope and at least 1 sputum specimen that is culture positive for M. tuberculosis

Finding suspects and sputum collection

Start Eligible adults visit the center

Symptom screening interview

Has a symptom and/or

abnormal CXR?

Collect sputum (2 times)

Stop Culture and microscopic

examination

Yes

No

Chest X-ray

Comparison of 2004 and 2012

Topic 2004 2012

Survey design Piggy back survey

of NSES Its own survey

Sampling design PSU: CB SSU: HH

PSU: Village SSU: Cluster

Stratification Urban/rural area Sumatera, Java-

Bali, Other

Sample size 16 HHs/CB

50,000 adults 500 adults/cluster

78,000 adults

Comparison of 2004 and 2012

Topic 2004 2012

Screening Symptom interview

Symptom interview and CXR

Case definition Smear Smear, culture, and

CXR

Data collection Survey team visits

HH Eligible adults visit

the center

Estimate TB prevalence of 2004 and 2012

If different, it may due to difference of Survey design and sampling strategy, including

the sample size

Screening method

Case definition

Data collection procedure

Need to have techniques to adjust the differences when comparison or trend analysis is necessary (?)

Thank you